Run the CIND indoubt tool from a terminal to test the effects
of indoubt failures on application programs. You can also use this
transaction to produce shunted units of work and change the characteristics
of units of work.
About this task
When active, the tool forces all transactions that are defined in the transaction class, DFHTCIND, to fail in doubt when they reach a synchronization point. You cannot use CIND on internal CICS® system transactions.
Procedure
- Change the transaction class of the transactions that you
want to go indoubt to DFHTCIND.
- Enter CIND ON. This command
activates the tool and enables a task-related user exit for the task
start. You get a message confirming that the tool
is active.
For every transaction that starts, the task-related user exit inquires on the transaction class. If the class is DFHTCIND, the task-related user exit issues a call to the CICS recovery manage to add a link to CIND, acting as a CICS coordinator system called DFHINDSP.
- Run you transactions. If the transactions update
recoverable resources and the TRANSACTION resource specifies the attribute
WAIT(YES), they fail in doubt. The transaction
fails with abend code ASP1 and the unit of work is shunted. You can
inquire on the shunted units of work.
A transaction that updates
nonrecoverable resources, or that issues only READ requests, fails
with abend code ASPR. The unit of work is not shunted because it is
read-only.
- Enter CIND OFF. This
command disables the task-related user exit and stops any transactions
going indoubt. You get a message confirming that
the tool is no longer active.
- Enter CIND RESYNC COMMIT or CIND
RESYNC BACKOUT. These commands simulate a coordinator CICS system reconnecting to your CICS region. The coordinator is DFHINDSP.
Results
The CICS recovery manager unshunts all units of work that were previously shunted because of the tool. The commit or backout specifies what action the recovery manager does when the unshunt takes place. For each unit of work that is unshunted, message DFHIN1012 is
sent to CSMT. This message details the unit of work status as defined
by DFHINDSP and the recovery manager.
You can also use these
commands to simulate what happens units of work are not synchronized.
For example, if a TRANSACTION resource specifies the attributes WAIT(NO)
and WAITACTION(COMMIT), the unit of work is not shunted but forced
to commit. If you enter a CIND RESYNC BACKOUT at
this point, the DFHIN1012 message output shows the
unit of work to be out of sync, as does message DFHRM0111.